videocut/cut_tail.py

182 lines
6.6 KiB
Python

import subprocess
import json
import os
import glob
import sys
def get_video_duration(video_path):
"""Uses ffprobe to extract the exact duration of a video."""
cmd = [
'ffprobe', '-v', 'error', '-show_entries', 'format=duration',
'-of', 'default=noprint_wrappers=1:nokey=1', video_path
]
result = subprocess.run(cmd, capture_output=True, text=True)
try:
return float(result.stdout.strip())
except ValueError:
return None
def get_previous_keyframe(input_video, cut_timestamp):
"""Scans backward to find the closest keyframe immediately PRECEDING the cut timestamp."""
# Look up to 5 minutes backward for a keyframe to be safe with heavy compression
start_search = max(0, cut_timestamp - 300)
cmd = [
'ffprobe', '-v', 'quiet', '-select_streams', 'v',
'-skip_frame', 'nokey', '-show_frames',
'-show_entries', 'frame=pkt_pts_time,pkt_dts_time,best_effort_timestamp_time',
'-of', 'json', '-read_intervals', f'{start_search}%{cut_timestamp}', input_video
]
result = subprocess.run(cmd, capture_output=True, text=True)
frames = json.loads(result.stdout).get('frames', [])
best_keyframe = 0.0 # Default to the start of the video if no keyframe is found
for frame in frames:
# Safely try multiple timestamp keys
time_str = (frame.get('best_effort_timestamp_time') or
frame.get('pkt_pts_time') or
frame.get('pkt_dts_time'))
if time_str is not None:
keyframe_time = float(time_str)
# Find the highest keyframe timestamp that is strictly less than or equal to our cut point
if keyframe_time <= cut_timestamp and keyframe_time > best_keyframe:
best_keyframe = keyframe_time
return best_keyframe
def smart_cut_tail(input_video, output_video, tail_duration):
"""Executes the Reverse Smart Cut process: Copy the bulk, re-encode the tiny tail tip, and stitch."""
# 1. Determine exactly where to cut
target_duration = get_video_duration(input_video)
if not target_duration:
print(f"Skipping {input_video}: Could not determine its total duration.")
return
cut_timestamp = target_duration - tail_duration
if cut_timestamp <= 0:
print(f"Skipping {input_video}: Video is shorter than the tail reference clip.")
return
print(f"Processing: {input_video} (Cutting tail at {cut_timestamp:.2f}s)...")
# 2. Find the preceding keyframe
prev_keyframe = get_previous_keyframe(input_video, cut_timestamp)
part1 = f"temp_part1_{input_video}"
part2 = f"temp_part2_{input_video}"
concat_list = f"concat_{input_video}.txt"
try:
# 3. Copy the bulk of the video (from 0 to the preceding keyframe)
if prev_keyframe > 0.0:
subprocess.run([
'ffmpeg', '-y', '-v', 'error', '-i', input_video,
'-t', str(prev_keyframe),
'-c:v', 'copy', '-c:a', 'copy', part1
], check=True)
# 4. Re-encode the tiny tip (from the preceding keyframe to the exact cut timestamp)
tiny_duration = cut_timestamp - prev_keyframe
subprocess.run([
'ffmpeg', '-y', '-v', 'error',
'-ss', str(prev_keyframe), '-i', input_video,
'-t', str(tiny_duration),
'-c:v', 'libx264', '-crf', '18', '-c:a', 'copy', part2
], check=True)
# 5. Concatenate them using a text list (Forced UTF-8 encoding for emojis/special chars)
with open(concat_list, 'w', encoding='utf-8') as f:
if prev_keyframe > 0.0:
f.write(f"file '{part1}'\n")
f.write(f"file '{part2}'\n")
subprocess.run([
'ffmpeg', '-y', '-v', 'error', '-f', 'concat', '-safe', '0',
'-i', concat_list, '-c', 'copy', output_video
], check=True)
print(f"Success! Saved to {output_video}")
except subprocess.CalledProcessError:
print(f"Error processing {input_video}. FFmpeg failed.")
finally:
# 6. Clean up temporary files safely
for temp_file in [part1, part2, concat_list]:
if os.path.exists(temp_file):
os.remove(temp_file)
def main():
if len(sys.argv) < 2:
print("Usage: python cut_tail.py <path_to_file_or_folder>")
return
# Get the absolute path and strip any accidental quotes
raw_input = sys.argv[1].strip(' "''')
input_arg = os.path.abspath(raw_input)
# Determine if it's a file or a folder
if os.path.isfile(input_arg):
if not input_arg.lower().endswith('.mp4'):
print("Error: The specified file is not an .mp4 video.")
return
target_dir = os.path.dirname(input_arg)
target_files = [os.path.basename(input_arg)]
elif os.path.isdir(input_arg):
target_dir = input_arg
target_files = None
else:
print(f"Error: Path '{input_arg}' does not exist.")
return
# Change the working directory to the target folder
os.chdir(target_dir)
# If it was a folder, grab all .mp4 files
if target_files is None:
target_files = glob.glob("*.mp4")
# WE NOW LOOK FOR 'tail_ref.mp4' INSTEAD OF 'baseline.mp4'
tail_ref_file = "o.mp4"
# Check if tail reference exists in the target directory
if not os.path.exists(tail_ref_file):
print(f"Error: '{tail_ref_file}' not found in the target folder: {target_dir}")
return
# Remove the reference file from the processing list so we don't try to cut it
if tail_ref_file in target_files:
target_files.remove(tail_ref_file)
if not target_files:
print("No other .mp4 files found to process.")
return
# Get exact duration of the tail reference
print(f"Analyzing tail reference duration: {tail_ref_file}...")
tail_duration = get_video_duration(tail_ref_file)
if not tail_duration:
print("Error: Could not determine the duration of the tail reference video.")
return
print(f"Tail reference duration found: {tail_duration} seconds.")
# Create output directory
output_dir = "processed_videos"
os.makedirs(output_dir, exist_ok=True)
print(f"\nFound {len(target_files)} video(s) to process in '{target_dir}'.\n" + "-"*30)
# Process each video
for video in target_files:
output_path = os.path.join(output_dir, video)
smart_cut_tail(video, output_path, tail_duration)
print("-" * 30 + "\nBatch processing complete!")
if __name__ == "__main__":
main()